Multi Level Guidance Of Urban Road Traffic Based On Congestion Prediction

As a negative gain product of rapid urbanization,traffic congestion has seriously affected the normal travel of citizens.The insufficient level of traffic management also makes the utilization rate of road resources low,and the unbalanced load of the road network aggravates urban congestion.In order to solve this drawback,many industry scholars at home and abroad have invested in research.Among them,route guidance has always been an important direction for the transportation industry to solve traffic congestion.Scientific and reasonable determination of the guidance space-time range and guidance action points is of great significance for improving the guidance efficiency.Based on this,this paper aims to solve the above-mentioned traffic congestion problem through the method of road traffic pre-control and guidance,and proposes a method for hierarchical guidance of nodes within the guidance range based on traffic congestion prediction,and verifies it with examples.First,this paper analyzes the main factors affecting traffic conditions,including traffic jam judgment factors and urban traffic jam characteristics.Taking into account the disadvantages of single parameter congestion recognition,an orderly decision model is selected to extract multiple features of congestion tags,and then optimized based on genetic algorithms The support vector machine classification model(GA-SVM)carried out a cross-combination experiment on an ordered set of congestion features,and finally determined three congestion feature indicators: flow rate,speed and the ratio of average travel time to free travel time.Then,use the machine learning recurrent neural network to construct a traffic congestion prediction model.Considering that the prediction data has the characteristics of a long-term sequence,a long and short-term memory neural network(LSTM)is selected as the traffic parameter prediction model,through data preprocessing and the use of spatio-temporal correlation The method determines the number of samples and the number of features in the model input data frame,which improves the prediction accuracy.Secondly,on the basis of the traffic state prediction model,by predicting the traffic state of the road network for a period of time in the future,it is judged whether the guidance conditions are met.For the congested road section that first meets the guidance conditions,it is determined based on the constraints of time and space Introduce the space-time range and construct a multi-level path guidance model;consider the difference in the function and capacity of each pre-control node in the guidance range,and determine the starting point set of guidance through priority screening of the pre-control nodes,and use the predicted information to make the A* shortest path The algorithm is improved to achieve multi-node multi-path guidance;considering the existence of road sections being shared by multiple guidance paths,the method of calculating the path scale is used to improve the relative utility of the Logit selection model,and the reasonable selection of the guidance path is realized.The reliability of the induced path is improved;finally,the load balance of the road network is realized through the dynamic loading method of batch traffic under the capacity limitFinally,take the regional road network in Chongqing with frequent congestion as an example,use Matlab and VISSIM to establish a simulation model,select SNSP,SNMP and T-RPIM strategies respectively for path search comparison experiments,and select DUA and SUA strategies for prediction information impact comparison experiments The results show that the method in this paper can effectively relieve urban traffic congestion,and the use of predictive information can effectively improve the induction efficiency.